Reciprocating high-pressure pumps (commonly called fracking, frac or well-service pumps) are often used in oil and gas fields for hydraulic fracturing of rock formations to increase hydrocarbon yields. Such pumps are typically truck-mounted for easy relocation from well-to-well. And they are usually designed in two sections: the (proximal) power section (herein “power end”) and the (distal) fluid section (herein “fluid end”). Each pump fluid end comprises at least one subassembly (and commonly three or more in a single fluid end housing), with each subassembly comprising a suction valve, a discharge valve, a plunger or piston, and a portion of (or substantially the entirety of) a pump fluid end subassembly housing (shortened herein to “pump housing” or “fluid end housing” or “housing” depending on the context).
For each pump fluid end subassembly, its fluid end housing comprises a pumping chamber in fluid communication with a suction bore, a discharge bore, and a piston/plunger bore. A suction valve (i.e., a check valve) within the suction bore, together with a discharge valve (i.e., another check valve) within the discharge bore, control bulk fluid movement from suction bore to discharge bore via the pumping chamber. Note that the term “check valve” as used herein refers to a valve in which a (relatively movable) valve body can reversibly close upon a (relatively stationary) valve seat to achieve substantially unidirectional bulk fluid flow through the valve.
Pulsatile fluid flow through the pump results from periodic pressurization of the pumping chamber by a reciprocating plunger or piston within the plunger/piston bore. Suction and pressure strokes alternately produce wide pressure swings in the pumping chamber (and across the suction and discharge check valves) as the reciprocating plunger or piston is driven by the pump power end.
Such pumps are rated at peak pumped-fluid pressures in current practice up to about 22,000 psi, while simultaneously being weight-limited due to the carrying capacity of the trucks on which they are mounted. (See, e.g., U.S. Pat. No. 7,513,759 B1, incorporated by reference).
Due to high peak pumped-fluid pressures, suction check valves experience particularly wide pressure variations between a suction stroke, when the valve opens, and a pressure stroke, when the valve closes. For example, during a pressure stroke with a rod load up to 350,000 pounds, a conventionally rigid/heavy check valve body may be driven longitudinally (by pressurized fluid proximal to it) toward metal-to-metal impact on a conventional frusto-conical valve seat at closing forces of about 50,000 to over 250,000 pounds (depending on valve dimensions). Total check valve-closure impact energy (i.e., the total kinetic energy of the moving valve body and fluid at valve seat impact) is thus converted to a short-duration high-amplitude valve-closure impulse (i.e., a mechanical shock). Repeated application of such a valve-closure shock with each pump cycle predisposes the check valve, and the fluid end housing in which it is installed, to vibration-induced (e.g., fatigue) damage. Cumulative shocks thus constitute a significant liability imposed on frac pump reliability, proportional in part to the rigidity and weight of the check valve body.
The emergence of new frac pump reliability issues (e.g., failures secondary to vibration-induced fatigue cracks) has paralleled the inexorable rise of peak pumped-fluid pressures in new fracking applications. And insight into these new pump failure modes can be gained through review of earlier shock and vibration studies, data from which are cited herein. For example, a recent treatise on the subject describes a mechanical shock “ . . . in terms of its inherent properties, in the time domain or in the frequency domain; and . . . in terms of the effect on structures when the shock acts as the excitation.” (see p. 20.5 of Harris' Shock and Vibration Handbook, Sixth Edition, ed. Allan G. Piersol and Thomas L. Paez, McGraw Hill (2010), hereinafter Harris). The above time and frequency domains are mathematically represented on opposite sides of equations generally termed Fourier transforms. And estimates of a shock's structural vibration effects are frequently described in terms of two parameters: (1) the structure's undamped natural frequency and (2) the fraction of critical structural damping or, equivalently, the resonant gain Q (see Harris pp. 7.6, 14.9-14.10, 20.10). (See also, e.g., U.S. Pat. No. 7,859,733 B2, incorporated by reference).
Mathematical representations of time and frequency domain data play important roles in computer-assisted analysis of mechanical shock. In addition, shock properties are also commonly represented graphically as time domain impulse plots (e.g., acceleration vs. time) and frequency domain vibration plots (e.g., spectrum amplitude vs. frequency). Such graphical presentations readily illustrate the shock effects of metal-to-metal valve-closure, wherein longitudinal movement of a check valve body is abruptly stopped by a valve seat. Relatively high acceleration values and broad vibration spectra are prominent, each valve-closure impulse response primarily representing a violent conversion of kinetic energy to other energy forms.
Since energy cannot be destroyed, and since a conventional valve can neither store nor convert (i.e., dissipate) more than a small fraction of the valve-closure impulse's kinetic energy, most of that energy is necessarily transmitted to the pump (or fluid end) housing. In a time domain plot, the transmitted energy appears as a high-amplitude impulse of short duration. And a corresponding frequency domain plot of transmitted energy reveals a broad-spectrum band of high-amplitude vibration. This means that nearly all of the check valve's cyclical valve-closure kinetic energy is converted to vibration energy. The overall effects of check valve closures may thus be compared to the mechanical shocks that would result from striking the valve seat repeatedly with a commercially-available impulse hammer, each hammer strike being followed by a rebound. Such hammers are easily configured to produce relatively broad-spectrum high-amplitude excitation (i.e., vibration) in an object struck by the hammer. (See, e.g., Introduction to Impulse Hammers at http://dytran.com/img/tech/all.pdf, and Harris p. 20.10).
Summarizing then, relatively broad-spectrum high-amplitude vibration predictably results from a typical high-energy valve-closure impulse. And frac pumps with conventionally-rigid valves can suffer hundreds of these impulses per minute. Note that the number of impulses per minute (for example, 300 impulses per minute) corresponds to pump plunger strokes or cycles, and this number may be converted to impulses-per-second (i.e., 300/60=5). The number 5 is sometimes termed a frequency because it is given the dimensions of cycles/second or Hertz (Hz). But the “frequency” thus attributed to pump cycles themselves differs from the spectrum of vibration frequencies resulting from each individual pump cycle impulse. The difference is that impulse-generated (e.g., valve-generated) vibration occurs in bursts of broad spectra which may simultaneously contain many vibration frequencies ranging from a few Hz to several thousand Hz (kHz).
Nearly all of this (generally-higher frequency) valve-generated vibration energy is quickly transmitted to proximate areas of the fluid end or pump housing, where it can be expected to excite damaging resonances that predispose the housing to fatigue failures. (See, e.g., U.S. Pat. No. 5,979,242, incorporated by reference). If, as expected, a natural resonance frequency of the housing coincides with a frequency within the valve-closure vibration spectrum, fluid end vibration amplitude may be substantially increased and the corresponding vibration fatigue damage made much worse. (See Harris, p. 1.3).
Opportunities to limit fluid end damage begin with experiment-based redesign to control vibration fatigue. For example, a spectrum of vibration frequencies initially applied to a fluid end as a test can reveal structural resonance frequencies likely to cause trouble. Specifically, the applied vibration of a half-sine shock pulse of duration one millisecond has predominant spectral content up to about 2 kHz (see Harris, p. 11.22), likely overlapping a plurality of fluid end housing natural frequencies. Such tests particularly focus attention on blocking progression of fatigue crack growth to the critical size for catastrophic fracture. Note that stronger housings aren't necessarily better in such cases, since increasing the housing's yield strength causes a corresponding decrease in critical crack size. (See Harris, p. 33.23).
It might be assumed that certain valve redesigns proposed in the past (including relatively lighter valve bodies) would have alleviated at least some of the above failure modes. (See, e.g., U.S. Pat. No. 7,222,837 B1, incorporated by reference). But such redesigns emerged (e.g., in 2005 or earlier) when fluid end peak pressures were generally substantially lower than they currently are. In relatively lower pressure applications (e.g., mud pumps), rigid/heavy valve bodies performed well because the valve-closure shocks and associated valve-generated vibration were less severe compared to shock and vibration experienced more recently in higher pressure applications (e.g., fracking). Thus, despite their apparent functional resemblance to impulse hammers, relatively rigid/heavy valves have been pressed into service as candidates for use in frac pump fluid ends. Indeed, they have generally been the only valves available in commercial quantities during the recent explosive expansion of well-service fracking operations. Substantially increased fluid end failure rates (due, e.g., to cracks near a suction valve seat deck) have been among the unfortunate, and unintended, consequences.
Under these circumstances, it is regrettable but understandable that published data on a modern 9-ton, 3000-hp well-service pump includes a warranty period measured in hours, with no warranty for valves or weld-repaired fluid ends.
Such baleful vibration-related results in fluid ends might usefully be compared with vibration-related problems seen during the transition from slow-turning two-cylinder automobile engines to higher-speed and higher-powered inline six-cylinder engines around the years 1903-1910. Important torsional-vibration failure modes suddenly became evident in new six-cylinder engines, though they were neither anticipated nor understood at the time. Whereas the earlier engines had been under-powered but relatively reliable, torsional crankshaft vibrations in the six-cylinder engines caused objectionable noise (“octaves of chatter from the quivering crankshaft”) and unexpected catastrophic failures (e.g., broken crankshafts). (Quotation cited on p. 13 of Royce and the Vibration Damper, Rolls-Royce Heritage Trust, 2003). Torsional-vibration was identified as the culprit and, though never entirely eliminated, was finally reduced to a relatively minor maintenance issue after several crankshaft redesigns and the development of crankshaft vibration dampers pioneered by Royce and Lanchester.
Reducing the current fluid end failure rates related to valve-generated vibration in frac pumps requires an analogous modern program of intensive study and specific design changes. The problem will be persistent because repeatedly-applied valve-closure energy impulses cannot be entirely eliminated in check-valve-based fluid end technology. So the valve-closing impulses must be modified, and their associated vibrations damped, to reduce excitation of destructive resonances in valves, pump housings, and related fluid end structures. Alternate materials, applied via innovative designs, illuminate the path forward now as they have in the past. Broad application of such improvements promises higher frac pump reliability, an important near-term goal. Simultaneously, inhibition of corrosion fatigue throughout analogous fluid circuits would be advanced, a longer-term benefit in refineries, hydrocarbon crackers and other industrial venues that are also subjected to shock-related vibration.